Date: Fri, 05 Oct 2012 11:45:04 +0300 From: Volodymyr Kostyrko <c.kworr@gmail.com> To: Martin Matuska <mm@FreeBSD.org> Cc: freebsd-fs@FreeBSD.org Subject: Re: [CFT] ZFS feature flag support for 9-STABLE Message-ID: <506E9E10.1010600@gmail.com> In-Reply-To: <506B4508.3030406@FreeBSD.org> References: <506B4508.3030406@FreeBSD.org>
next in thread | previous in thread | raw e-mail | index | archive | help
02.10.2012 22:48, Martin Matuska wrote: > ZFS feature flag support is ready to be merged to 9-STABLE. > The scheduled merge date is short after 9.1-RELEASE. > Feedback and suggestions are welcome. +1 from me: works here like a charm. I have one more question about snapshot removal behavior. I already noticed that destroying a snapshot an deduped filesystem can eat a lot of ram to complete. Actually my experience was: 1. One of the snapshots was discarded (14G) and system tried to allocate more RAM then was available. Machine stalled. 2. After reboot snapshot destruction was automatically restarted. And machine stalled again. 3. I booted in single user mode and tried a manual pool import. RAM was exhausted again. I assumed that adding swap wouldn't help either. 4. I raised memory on the machine from 4G to 8G. After boot snapshot destruction completes topping wired at 6G. As I understand snapshot destruction goes into one transaction meaning that all changed structures should be changed in memory and written to disk. Would this code have the same issue? -- Sphinx of black quartz, judge my vow.
Want to link to this message? Use this URL: <https://mail-archive.FreeBSD.org/cgi/mid.cgi?506E9E10.1010600>